#
#asahi-dev · 2025-12-16 12:15
chadmedi should take a look at direct scanout and the planar yuv formats again
#asahi-dev · 2025-12-16 12:12
oliver_thanks guys, i'll have a look tomorrow and will adjust the constants. i'll probably have some questions regarding the mmio regions.
#asahi-dev · 2025-12-16 12:09
chaos_princessnot even pll but a simple scale
#asahi-dev · 2025-12-16 12:09
chaos_princessi think so
#asahi-dev · 2025-12-16 12:09
chadmedthat is, the dcp's timebase will still be 24mhz on those machines
#asahi-dev · 2025-12-16 12:08
chadmedthe 1ghz timer in m3+ is just PLLed from the real 24mhz clock right
#asahi-dev · 2025-12-16 12:05
chadmedalmost as bad as tracing sep keybag
#asahi-dev · 2025-12-16 12:05
chaos_princessyes
#asahi-dev · 2025-12-16 12:04
oliver_chaos_princess: the test you've proposed is to have 120hz content on screen on the mac?
#asahi-dev · 2025-12-16 12:04
chadmedbut waiting for swaps while tracing sucks
#asahi-dev · 2025-12-16 12:04
chadmedwell yeah thats gonna be the only way to be certain
#asahi-dev · 2025-12-16 12:04
chaos_princessor do the test i proposed :P
#asahi-dev · 2025-12-16 12:03
chadmed(e.g. 120hz should be 8ish ms)
#asahi-dev · 2025-12-16 12:03
chaos_princessso, you are running at 120hz, but with 60hz worth of latency
#asahi-dev · 2025-12-16 12:03
chaos_princesssomething tells me ts1 is the "desired presentation time"
#asahi-dev · 2025-12-16 12:03
chadmedsee what happens there
#asahi-dev · 2025-12-16 12:03
chadmedi would try setting the ts offsets based on the current mode's refresh rate
#asahi-dev · 2025-12-16 12:03
oliver_so i am halfway confident that it actually updates with 120hz
#asahi-dev · 2025-12-16 12:03
oliver_and i have a second laptop with a 120hz screen here where i could reproduce the results of the test programm on a non-macbook
#asahi-dev · 2025-12-16 12:02
oliver_chadmed: yea using the demo programm from the pr as j`ey said.
#asahi-dev · 2025-12-16 12:02
chadmedalso yes 16.666 ms between present and ts3 screams 60hz
#asahi-dev · 2025-12-16 12:02
j`eyshould draw 4 balls or just 2 depending on 60 or 120 hz
#asahi-dev · 2025-12-16 12:01
chaos_princess(something tells me the offsets will change)
#asahi-dev · 2025-12-16 12:01
j`eychadmed: there's a test mentioned in the PR
#asahi-dev · 2025-12-16 12:01
chaos_princesstry making sure you have 120hz content on the screen
#asahi-dev · 2025-12-16 12:01
chadmedmacos cannot run at a fixed 120hz its VRR or 60hz
#asahi-dev · 2025-12-16 12:01
chaos_princesssince macos only runs at 120hz when something wants it
#asahi-dev · 2025-12-16 12:00
chadmedalso, did you confirm that the screen is indeed swapping at 120hz on linux?
#asahi-dev · 2025-12-16 12:00
chaos_princessits probably running at 60hz at that point
#asahi-dev · 2025-12-16 12:00
chadmedyeah tracing dcp is painful
#asahi-dev · 2025-12-16 12:00
oliver_login screen, maybe desktop already. needed to move things around to get submits.
#asahi-dev · 2025-12-16 12:00
chaos_princessi mean in macos
#asahi-dev · 2025-12-16 11:59
chadmedtry spamming ctm or brightness changes :P
#asahi-dev · 2025-12-16 11:59
chaos_princessoliver_: what was on screen when you got those values?
#asahi-dev · 2025-12-16 11:58
chaos_princessso, that is probably fine?
#asahi-dev · 2025-12-16 11:58
chaos_princess1s/16ms is 60-ish
#asahi-dev · 2025-12-16 11:57
chadmedits also dog slow on like, every machien
#asahi-dev · 2025-12-16 11:57
chadmediomfb_flush is time sensitive so idk if experimental constants are gonna fly here tbh
#asahi-dev · 2025-12-16 11:57
oliver_yea, i've just graphed the dcp trace values and looked for anything that looked like a pattern.
#asahi-dev · 2025-12-16 11:56
chadmedexperimentally determined by the looks, its in the commit message
#asahi-dev · 2025-12-16 11:56
chaos_princessi assume ts1 is present time and ts2 is render completion time?
#asahi-dev · 2025-12-16 11:56
chaos_princessalso, where did 16660000 and 40000000 come from
#asahi-dev · 2025-12-16 11:55
chaos_princesstimesync? no, but it is just yet another mmio on the dcp, look for the smallest one, iirc it was that one
#asahi-dev · 2025-12-16 11:55
chadmedi.e ns_to_mach_time()
#asahi-dev · 2025-12-16 11:55
chadmedoliver_: also nit but s/convert//g
#asahi-dev · 2025-12-16 11:55
oliver_chaos_princess: i see. is that already in the kernel? then i can have a look into that the the next few days
#asahi-dev · 2025-12-16 11:54
chadmedbut yes this should just be plumbed into the standard drm apis for mode setting
#asahi-dev · 2025-12-16 11:54
chaos_princessin theory you want proper vrr, the display can go as low as 10 hz
#asahi-dev · 2025-12-16 11:54
oliver_j`ey: you can just select 60hz I guess
#asahi-dev · 2025-12-16 11:54
chaos_princess^

plus ›